A Marine Debris Detection Framework for Ocean Robots via Self-Attention Enhancement and Feature Interaction Optimization

TL;DR

This paper introduces YOLO-MD, a novel marine debris detection framework for ocean robots that enhances feature extraction and addresses class imbalance, achieving superior accuracy in degraded images.

Contribution

The paper presents a dual-branch self-attention module, a shift-based operation, and a dynamic reweighting loss to improve marine debris detection in challenging conditions.

Findings

Achieved 0.875 precision and 0.849 mAP50 on UODM dataset.

Outperformed state-of-the-art methods in marine debris detection.

Validated effectiveness through real-world robotic deployment.

Abstract

Marine debris detection for ocean robot is crucial for ecological protection, yet performance is often degraded by low-quality images with blur, complex backgrounds, and small targets. To address these challenges, we propose YOLO-MD, an enhanced YOLO-based detection framework. A Dual-Branch Convolutional Enhanced Self-Attention (DB-CASA) module is designed to strengthen spatial-channel interactions, improving feature representation in degraded images. Additionally, a lightweight shift-based operation is introduced to enhance fine-grained feature extraction for objects of varying scales while maintaining parameter efficiency. We further propose SFG-Loss to mitigate class imbalance and optimization instability via dynamic sample reweighting. Experiments on the UODM dataset demonstrate that YOLO-MD achieves 0.875 precision, 0.822 F1-score, and 0.849 mAP50, outperforming the latest state-of-the-art methods. The effectiveness of this method has also been verified through real-world robotic edge deployment experiments.